D\'ecouvrir de nouvelles classes dans des donn\'ees tabulaires

TL;DR

This paper introduces TabularNCD, a novel method for discovering new classes in heterogeneous tabular data by leveraging known classes, pseudo labels, and multi-task learning, extending NCD beyond image data.

Contribution

The paper proposes a new framework, TabularNCD, specifically designed for heterogeneous tabular data, including a novel pseudo label method and joint optimization strategy.

Findings

Demonstrates NCD applicability to tabular data

Introduces a new pseudo label method for heterogeneous variables

Shows improved class discovery in tabular datasets

Abstract

In Novel Class Discovery (NCD), the goal is to find new classes in an unlabeled set given a labeled set of known but different classes. While NCD has recently gained attention from the community, no framework has yet been proposed for heterogeneous tabular data, despite being a very common representation of data. In this paper, we propose TabularNCD, a new method for discovering novel classes in tabular data. We show a way to extract knowledge from already known classes to guide the discovery process of novel classes in the context of tabular data which contains heterogeneous variables. A part of this process is done by a new method for defining pseudo labels, and we follow recent findings in Multi-Task Learning to optimize a joint objective function. Our method demonstrates that NCD is not only applicable to images but also to heterogeneous tabular data.